Scholarlink

<p>The accelerating evolution of cyber threats presents a critical challenge to the security and resilience of high-speed industrial communication networks. Traditional defense mechanisms often struggle to keep up with the dynamic and sophisticated nature of these threats. To overcome this limitation, we propose a novel <b>L</b>ong <b>S</b>hort-<b>T</b>erm <b>M</b>emory-integrated <b>T</b>ransfer <b>L</b>earning-based <b>C</b>yber <b>T</b>hreat <b>I</b>ntelligence model (<b>LSTM-TL-CTI</b>). This model adopts a two-tier architecture that combines adaptive learning and real-time intelligence sharing to enhance threat detection and response capabilities. In the first tier, LSTM networks are employed to proactively detect anomalies by capturing temporal patterns in local network traffic. The second tier uses transfer learning to share the learned knowledge between distributed network segments, enabling continuous model refinement and scalable cyber threat intelligence. This collaborative learning strategy ensures the rapid detection of both known and emerging threats, leading to robust and adaptive defense mechanisms. Extensive test-bed simulations using real world benchmark datasets validate the effectiveness of the proposed framework. The LSTM-TL-CTI model achieves up to 80.29% improvement in detection accuracy and increases true positive rates by 64.25% compared to conventional neural network approaches. These results highlight the potential of the model as a high-impact solution to protect modern industrial communication networks against evolving cyber threats.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An adaptive cyber threat intelligence model to counter evolving security attacks in industrial communication networks

  • 期刊论文

  • 发表日期：

    2026年1月23日

查看全文
  • Randima Nimantha Wedamuni Arachchige,
  • Deepika Saxena,
  • Ashutosh Kumar Singh

摘要

The accelerating evolution of cyber threats presents a critical challenge to the security and resilience of high-speed industrial communication networks. Traditional defense mechanisms often struggle to keep up with the dynamic and sophisticated nature of these threats. To overcome this limitation, we propose a novel Long Short-Term Memory-integrated Transfer Learning-based Cyber Threat Intelligence model (LSTM-TL-CTI). This model adopts a two-tier architecture that combines adaptive learning and real-time intelligence sharing to enhance threat detection and response capabilities. In the first tier, LSTM networks are employed to proactively detect anomalies by capturing temporal patterns in local network traffic. The second tier uses transfer learning to share the learned knowledge between distributed network segments, enabling continuous model refinement and scalable cyber threat intelligence. This collaborative learning strategy ensures the rapid detection of both known and emerging threats, leading to robust and adaptive defense mechanisms. Extensive test-bed simulations using real world benchmark datasets validate the effectiveness of the proposed framework. The LSTM-TL-CTI model achieves up to 80.29% improvement in detection accuracy and increases true positive rates by 64.25% compared to conventional neural network approaches. These results highlight the potential of the model as a high-impact solution to protect modern industrial communication networks against evolving cyber threats.